Abstract: The characteristics of condensational heating rates of both grid -scale and subgrid-scale in Shijiazhuang torrential rain event happened in Aug. 3-4 of 1996 are investigated using real data numerical simulation results from 4 experiments of MM5 model (△x=20 km), in which 4 convective parameteriz ation schemes, KUO, GRELL, KAINFRITSCH and BETTS-MILLER (here after refers as to K U, GR, KF, and BM) are chosen respectively. And the impacts of condensational he ating rates on mesoscale structures of the torrential rain simulations are also studied.It is found that the distributions of subgrid-scale condensational heating rates of 4 runs are similar in the vertical heating peak feature (at the middle level of the troposphere), but are different in the heating thickness and strength; w hereas those of the grid-scale from 4 experiments are quite similar in heating thickness with heating peak at the low-level of the troposphere, and there are some differences in the heating strengths too. In addition, the distributions of total condensational heating rates in GR, KF and BM runs are determined by their grid-scale heating rates, while in KU run, it is controlled by subgrid-scale heating rates. Despite the differences of the total condensational heating rates between the 4 runs in the period of the torrential rain are quite limit, their impacts on the mesoscale structures in the simulations are negligible. The differences of conde nsational total heating rates affect the mesoscale structures and evolutions of the torrential rain in a way of chain reaction, namely the differences of the total heating rates induce the discrepancies on mesoscale circulation over the rai ning region, and these local circulation discrepancies cause further the diversi ties of the precipitation simulations.It seems that the KF run is more reasonable in simulating the middle-latitude s ystems than others; whereas it is easy to occur in the KU run that the moisture and temperature of the grid column are adjusted excessively and unreasonably by the convective parameterization at 20 km or higher horizontal resolutions. And for a reliable mesoscale simulation, it is very important to verify the detail st ructures and time series of precipitation simulations against observations beca use the detail structures and evolutions of the precipitation are closely relate d with the distributions and evolutions of the condensational heating and mesosc ale circulations.